The “Bad Likert Judge” jailbreaking technique boasts a high attack success rate by using a three-step approach which employs the target LLM’s own understanding of harmful content to bypass the target LLM’s safety guardrails....more
Model inversion and membership inference attacks create unique risks to organizations that are allowing artificial intelligences to be trained using their data. Companies may wish to begin to evaluate ways to mitigate risks...more
New and old attack vectors analyzed by RAND in their report on securing AI weights from theft. A new report published by RAND highlights the importance of securing the learnable parameters, or weights, of AI models to...more
California continues as the frontrunner of U.S. AI regulatory developments. Following the Governor’s executive order on Generative AI (GenAI) published last year, California state agencies have worked to implement its...more